AMD plans silicon fix for Spectre vulnerability

Comeback kid AMD announced on its quarterly earnings call that it intends to have a silicon fix for the variant 2 of the Spectre exploit, the only one of the Meltdown and Spectre exploits it’s vulnerable to, by 2019 with its new Zen 2 core.

The company also said it will ramp up GPU card production to meet the insane demand these days thanks to cryptominers, although it said the biggest challenge will be to find enough memory to make the cards.

It’s hard to believe that in 2018 we are seeing such shortages in computing hardware, but there you have it.

It was an upbeat earnings call for AMD for once. Fourth quarter revenues came in at $1.48 billion and non-GAAP EPS came in at $0.08 per share, beating estimates for $1.41 billion and $0.05, respectively. Revenue grew 33 percent over the prior-year period, and in that same period the company saw a penny per share loss. So, the new Zen core is bringing some bliss to AMD’s finance department.

AMD had significantly lower exposure to the Meltdown and Spectre vulnerabilities than Intel because it did not use the branch prediction method Intel uses and where the exploit lies. It’s only real vulnerability was to variant 2 of Spectre, and even there the company says risk is minimal.

Nevertheless, CEO Lisa Su told analysts on the earnings call, “We continue to believe that Variant 2 of Spectre is difficult to exploit on AMD processors. However, we are deploying CPU microcode patches that in combination with OS updates provide additional mitigation steps. Longer term, we have included changes in our future processor cores, starting with our Zen 2 design, to further address potential Spectre-like exploits.”

The current core generation that has turned AMD around is known as Zen. This year, the company is believed to release a Zen+ core with new chips, focused primarily on performance improvements, which should be something because Zen already performs incredibly well.

Zen 2 — A die shrink and a new core

Zen 2 is scheduled for 2019 and is set to accompany a die shrink to 7nm. This makes me nervous because Intel has been bedeviled with trying to get down to 10nm, and if Intel struggles with 10nm, what is Globalfoundries, AMD’s chip manufacturing partner, going to go through getting to 7nm? Plus, a die shrink and new core at the same time smacks of too much risk at once, and I don’t want to see a repeat of the Barcelona fiasco.

AMD CTO Mark Papermaster, who is no slouch in the CPU design space (IBM and Apple once got into a legal brawl over him) told EE Times last year that the company had to “double” its efforts “across foundry and design teams” in the build up to 7nm. He said it’s been the toughest generational bump he has seen in a long time, requiring completely new CAD tools and changes to architecture.

Separately, the company said it will attempt to ramp up GPU production if memory supplies will allow it. GPUs are being gobbled up by cryptominers looking for Bitcoin and other cryptocurrencies because GPUs are far more efficient at mining the coins than CPUs. The result has been a shortage so bad Nvidia has actually asked retailers not to sell more than two cards at a time to a customer and try and ensure the cards go to gamers, not miners.

For AMD, the challenge is a lack of memory on the market. During the earnings call, Su said, “At this point we are not limited by silicon per se, so our foundry partners are supplying us, there are shortages in memory and I think that is true across the board, whether you are talking about GDDR5, or you’re talking about high bandwidth memory. We continue to work through that, with our memory partners and that will be certainly one of the key factors as we go through 2018.”

CPU exploits, GPU shortages, memory shortages … who said hardware is boring?